When the soil was shoveled loosely into the tank, without consolidation of any kind, it settled on becoming saturated and became quite compact. It took five days for the water to appear in the sixth gauge pipe near the lower end of the tank. After the pressure, which was maintained constant, had been on for several weeks, the seepage amounted to one gallon in 22 minutes. When the soil was deposited by shoveling into the water, the seepage amounted to one gallon in 34 minutes. The relative filtering capacities of soils and sands were thought to be better determined by the use of galvanized iron cylinders of known areas. Fig. 23 shows one of the cylinders. These latter experiments confirmed those previously made at Lawrence, by Mr. Allen Hazen, for the Massachusetts State Board of Health. They showed that the loss of head was directly proportioned to the quantity of water filtered and that the quantity filtered will vary as the square of the diameter of the _effective size_ of the grains of the filtering material.[4] The material classed as “permeable” at the North Dike of the Wachusett Reservoir has an effective diameter of about 0.20 mm. A few results are given in the following table: Amount of Filtration in Gallons per Day, Through an Area of 10,000 Sq. Ft., With a Loss of Head or Slope of 1 ft. in 10 ft. Material. Unit ratios. U. S. gallons. (1) Soil 1 510 (2) Very fine sand 14 7,200 (3) Fine sand 176 90,000 (4) Medium sand 784 400,000 (5) Coarse sand 4,353 2,200,000 To be sure that the accumulation of air in the small interstices of the _soil_ was not the cause of the greatly reduced filtration through it, another series of experiments was conducted in the wooden tank, as shown in Fig. 24. A pair of screens was placed near each end of the tank, filled with porous material, sand and gravel, and the 50-ft. space between filled with soil. The soil was rammed in 3-in. layers, and special care taken to prevent water from following along the sides and bottom of the tank. One end was filled with water to near the top, while the other end gave a free outlet. After this experiment had been continued for more than a month, the amount of seepage averaged 1.7 gallons per 24 hours, or about 32 drops per minute. Filtration tests were also made through soil under 150 ft. head, or 5 lbs. per sq. in., with results not materially different, it is stated, from those already given. The soil used in all these tests contained from 4 to 8% by weight of organic matter. This was burned and similar tests made with the incinerated soil, resulting in an increase of about 20% more seepage water.